This invention relates to methods for the synthesis of monocrystalline films and crystals that are suitable as substrates to support the growth of cubic monocrystalline 3C-silicon carbide (.beta.-SiC).
It has long been believed that monocrystalline .beta.-SiC would be a superior semiconductor material to silicon (Si). Monocrystalline .beta.-SiC has no polytypic form and its electronic transport properties are superior to Si and all other SiC forms. The superior properties can potentially extend the range of solid state electronic applications beyond present power, power-frequency, temperature and radiation density limits. .beta.-SiC seems particularly suited for operation at high temperature such as, for example, 400.degree.-600.degree. C.
SiC was early identified as a candidate material for integrated circuitry applications since it has a high breakdown voltage, relatively large band gap, and a thermal conductivity of more than three times that of silicon at ambient temperature. SiC is also resistant to the diffusion of impurity species. It has been hoped that SiC devices can be substituted for Si devices.
Despite its apparent advantages, SiC has not been used for semiconductor devices because it has not been possible to produce unpolytyped SiC single crystals of sufficient size to allow the fabrication of semiconductor devices. To be useful, a .beta.-SiC single crystal must be substantially free of defects. But, to date, there are no reliable techniques to produce defect-free .beta.-SiC single crystals of a sufficient size to be used in integrated circuit production.
Nucleation of the .beta. form of SiC occurs below 1700.degree. C. and useful bulk SiC crystal growth rates occur at much higher temperatures. Since SiC is only thermodynamically stable in an .alpha.-SiC phase at temperatures exceeding 1600.degree. C. there has been no practical way to directly grow monocrystalline .beta.-SiC from a melt or by sublimation. Small bulk single crystals of .beta.-SiC have been fabricated, but with dimensions no greater than a few millimeters.
The only practical way to synthesize high-purity .beta.-SiC single crystals appears to be by thin film growth from a vapor at temperatures below 1600.degree. C., such as by chemical vapor deposition (CVD), or gas source molecular beam epitaxy. Several methods have been used to transport Si and C atoms to a substrate for the purpose of depositing .beta.-SiC as a thin film. Further, several methods have been employed to initiate nucleation and growth, and to sustain growth of .beta.-SiC. Examples of such methods appear in U.S. Pat. Nos. 4,767,666 and 4,923,716.
Epitaxy has not yet worked as hoped due to the lack of a suitable monocrystalline substrate to support the heteroepitaxial growth of monocrystalline .beta.-SiC. To be successful, a substrate would need to have certain properties with respect to .beta.-SiC. First, the thermal coefficient of expansion of the substrate should be greater than (to compressively load the epilayer) or equal (to minimize strain) to that of .beta.-SiC. Second, any lattice parameter mismatch between the substrate and the .beta.-SiC should be less or about equal to 1%. Third, the substrate must be thermally and chemically stable under conditions required for .beta.-SiC growth. Also, for cost efficiency it would be desirable if the substrate could be fabricated in bulk, single crystal form.
The most important part of a substrate for single-crystal epitaxial growth of thin film is its surface, upon which the thin film is nucleated and grown. In general, substrates are cut from bulk crystals and the surface of the substrate is polished to a flat smooth surface. The reason that the entire substrate must be a single crystal with low defect concentration is that its surface properties will reflect its bulk properties. Thus, if a substrate contains a high defect density or it is polycrystalline, then its surface will have a high defect density or it will be polycrystalline.
Unfortunately, each of the previously used substrates has lacked one of the above-mentioned properties.
The substrates which have been used for the purpose of attempting to nucleate monocrystalline, epitaxial layers of .beta.-SiC are: Al.sub.2 O.sub.3, AlN, 6H.alpha.-SiC, Si, and TiC. There have also been a few reports of .beta.-SiC epitaxy on very small .beta.-SiC substrates.
The development of 6H.alpha.-SiC bulk crystal growth capabilities spurred interest in its use as a substrate for .beta.-SiC epitaxial growth. However, it has been found that 6H.alpha.-SiC substrate surfaces do not have suitable properties for nucleating low-defect concentration, monocrystalline .beta.-SiC thin films. In particular, experimental evidence shows that .beta.-SiC can only be nucleated on the perfectly oriented (0001) and (0114) 6H.alpha.-SiC surfaces, where &lt;111&gt; .beta.-SiC is nucleated on (0001) 6H.alpha.-SiC surfaces, and &lt;001&gt; .beta.-SiC is nucleated on (0114) 6H.alpha.-SiC surfaces. The slightest misorientation results in homoepitaxial growth of 6H.alpha.-SiC.
Experimental evidence also shows that .beta.-SiC epilayers grown on the (0001) 6H.alpha.-SiC surfaces contain double positioning boundaries (DPB) defects which make them useless for semiconductor device and IC development. DPB defects can be explained as follows. There are six atoms on the surface of the hexagonal basal plane to which depositing atoms can bond. If the depositing atoms nucleate as a cubic structure, then there can be only three bonds attached to the six surface plane atoms. Thus, to form a cubic structure, the nucleating atoms must attach to every other atom in the basal plane of the hexagonal lattice. This causes the random nucleation of two .beta.-SiC epilayer orientations. Both orientations are identical perpendicular to the 6H.alpha.-SiC substrate, but they are misoriented 60.degree. with respect to each other in the plane of the substrate surface. The DPBs between these two orientations are incoherent and they have a high internal energy. The internal energy of the DPBs is released by the formation of stacking faults. These stacking faults seriously degrade the electronic transport properties of the .beta.-SiC epilayer.
In attempts to grow .beta.-SiC on Si substrates, the resulting .beta.-SiC crystal has been found to contain high concentrations of crystallographic defects resulting from significant mismatches between the lattice parameters and the expansion coefficients of the two materials.
Of the substrates tested, only TiC and .beta.-SiC have ever yielded monocrystalline, epitaxial .beta.-SiC substantially free of microcracks, antiphase domains, and double positioning boundaries. The problem with these substrates is that they are difficult to synthesize as large single crystals. The development of .beta.-SiC single-crystal growth technology, where the .beta.-SiC crystals are of significant size, may not be possible at all because of the relationship between silicon carbide crystalline form and the temperature at which it is synthesized.
TiC is particularly suitable as a substrate for nucleation and growth of monocrystalline, epitaxial .beta.-SiC. The lattice parameter mismatch to .beta.-SiC is less than 1% and TiC forms an almost ideal Schottky interface with .beta.-SiC, which is beneficial for ohmic contacts to n-type .beta.-SiC in vertical devices. Because both materials are cubic, the problem of double positioning boundaries can be completely eliminated. TiC and .beta.-SiC have similar thermal expansion, and titanium is electrically inert in .beta.-SiC. However, TiC has not been a successful substrate due to the difficulty of producing TiC single crystals possessing the requisite quality. Prior TiC substrates have suffered from pinholes and subgrain boundary defects. It has been found that the pinholes and the subgrain boundaries in the TiC substrate generate the same defects into the structure of .beta.-SiC film grown on the TiC.
For this reason, there is still a need for a substrate on which one can successfully grow monocrystalline .beta.-SiC that is substantially defect free and of a useful size.